Changing the source of existing directlake model

[Spiedo]

Hi Data friends,

I am facing the following problem. how can i easily change the source of a directlake table which has the same attributes but different table name? 

I have a "directlake semantic model" within a fabric enabled workspace. 

I modeled around 14 tables and for some tables the source name will change while keeping the same attributes. When i want to change the source tables using tabular editor or management studio , it looks like the change happened and has been applied. But when i click refresh in the web-browser the old source is still being used by the directlake semantic model.

Within management studio i get this response :

Executing the query ...
READS, 0
READ_KB, 0
WRITES, 0
WRITE_KB, 0
CPU_TIME_MS, 0
ROWS_SCANNED, 0
ROWS_RETURNED, 0

Lists of BinaryXml value tokens not supported.
Run complete

So it does not seem to work. In this case i ran XMLA, created it with "create or replace" scripting option from management studio and altered the entityname part into the desired tablename

"partitions": [
{
"name": "Audit Informatie AI",
"mode": "directLake",
"source": {
"type": "entity",
"entityName": "dim_audit",
"expressionSource": "DatabaseQuery",
"schemaName": "dlr"
}

into

{
"name": "Audit Informatie AI",
"mode": "directLake",
"source": {
"type": "entity",
"entityName": "DimAudit",
"expressionSource": "DatabaseQuery",
"schemaName": "dlr"
}


And the entity does exist in the same lakehouse as the dim_audit

[Spiedo · Accepted solution]

Hi Rohit

Thanks for the help. I could not find the alter database in SQL management studio as this is an analysis model and doesn't have the scripting option. Did not find the  refresh>clear cache, possibly because caching is turned off on our client. 

But my colleague helped me . Instead of scripting the partition into TSML using tabular editor and changing the table name, the whole table needs to be scripted using tsml and one should edit the source lineagetable. Then it does work 🙂
I fired the TSML in a MDX query pane which resulted in the desired outcome

[rohit1991]

Hi @Spiedo ,

If the new table name isn't reflecting in your DirectLake Semantic Model, follow these steps:

1. Use XMLA with Full Update

Instead of using "create or replace", try "Alter Database" in Management Studio. This ensures the model structure updates properly.

2. Clear DirectLake Cache

Power BI caches DirectLake models, which may prevent updates from applying.

• In Power BI Service, go to the Dataset > Refresh > Clear Cache option.
• This forces Power BI to reload the table structure from the new source.

3. Manually Refresh in Fabric Workspace

• Open your Fabric workspace where the model exists.
• Select the Semantic Model > Refresh in the Power BI Web UI.
• This helps sync any metadata changes.

4. Verify Table Permissions in OneLake

Check that the new table name (DimAudit) has the same access permissions as the previous table. If permissions are different, the update may not apply correctly.

This should ensure that your DirectLake model properly switches to the new source table. Let me know if you need further clarification.